Research on CNN-Attention Regression Prediction Method Based on GOA Optimization

In recent times, deep learning technology has made remarkable strides across diverse domains. The Convolutional Neural Network (CNN) is widely applied in various domains, including image recognition and natural language processing, showcasing its versatility and significance within the realm of deep learning. In the problem of regression prediction, the introduction of CNN-Attention model has also achieved some results. This paper proposes a CNN-attention regression prediction method optimized by locust algorithm, which combines the feature extraction capability of CNN and the attention-weighting function of the Attention mechanism. In contrast to the conventional CNN model, The CNN-Attention model presented in this work incorporates an Attention mechanism within its fully connected layer, enabling dynamic adjustment of the significance of each feature, thereby enhancing prediction efficiency and generalization capability. Additionally, the locust algorithm is employed to refine the CNN-Attention model. By simulating the search behavior of locust population, the model parameters are optimized to improve the performance and convergence speed of the model. Experimental outcomes indicate that the CNN-Attention model, fine-tuned using the locust algorithm, has delivered commendable performance in the regression prediction task, exhibiting high precision and robustness, and provides an effective deep learning solution for solving the regression prediction problem in practical applications.